Exploring Word Order Predictions of Theories of Communicatively Efficient Language Production

## 
## Assigned RegType
## inferWeightsCross.py
## readDataDistCrossGPUFreeMomentumEarlyStop.py
## readDataDistCrossGPUFreeMomentumEarlyStopEntropyPersevereAnnealAllCorpora.py
## readDataDistCrossGPUFreeMomentumEarlyStopEntropyPersevereAnnealAllCorporaOnlyLexRestrNP.py
## inferWeightsCrossVariationalAllCorpora.py
## readDataGradientsWordSurp.py
## readDataDistEnglishGPUOnlyDependencyLabelNoPred.py
## readDataDistEnglishGPUFreeUIDCorrectMean.py
## readDataDistCrossLGPUDepLengthMomentumEntropyUnbiasedBaseline.py
## readDataDistEnglishGPUFreeMomentum.py
## readDataDistCrossLCPUDepLengthMomentum.py
## readDataDistCrossGPUFreeMomentum.py
## readDataDistCrossGPUFreeMomentumEarlyStopEntropyPersevere.py
## readDataDistCrossGPUFreeMomentumEarlyStopEntropy.py
## readDataDistCrossGPUFreeMomentumEarlyStopEntropyPersevereAnneal.py
## readDataDistCrossLGPUDepLengthMomentumEntropy.py
## readDataGradientsDepL.py
## readDataDistCrossGPUFreeMomentumEarlyStopEntropyPersevereAnnealAllCorporaOnlyLex.py
## readDataDistEnglishGPUFreeUIDLagrangeRelaxPrimalDualPOS.py
## readDataDistEnglishGPUFreeParsingUnbiased.py
## readDataDistCrossLGPUDepLengthMomentum.py
## readDataDistEnglishGPUFreeParsingUnidir.py
## readDataDistEnglishGPUFreeUID.py
## readDataDistCrossLGPUDepLengthMomentumEntropyPlainBaseline.py
## inferWeightsCrossVariational.py
## readDataDistEnglishGPUFreeUIDCorrectMeanAllCorporaRate.py
## inferWeightsForEnglish.py
## readDataDistEnglishGPUDepLength.py
## readDataDistEnglishGPUFreeUIDLagrangeRelax.py
## readDataDistCrossGPUFreeMomentumEarlyStopEntropyLR.py
## readDataDistEnglishGPUFreeUIDLagrangeRelaxPrimalDual.py
## readDataDistEnglishGPUFreeParsingUnbiasedEntropy.py
## readDataRegressionDepLengthAndSurprisalRandomEffects.py
## readDataDistEnglishGPUFreeUIDCorrectMeanAllCorpora.py
## readDataDistCrossGPUFreeMomentumEarlyStopEntropyPersevereAnnealSmallVocab.py
## readDataRegressionDepLengthAndSurprisalRandomEffectsVariationalUIDRandIntercept.py
## readDataDistCrossGPUFreeMomentumEarlyStopGaussianPersevere.py
## readDataGradientsUIDs.py
## Take into account (1) log variances, (2) UID rate vs UID TODO do same with other variances and SigmasALSO take into account depLogLengthWhere is Norwegian surprisal data???

Slopes for Dependency Length (by dependency triple)

Here, I show fitted per-dependency-triple slopes for dependency length, first by language and then for all languages together. A more negative slope indicates above-average effect of dependency length. Only the 15 most frequent triples are shown per language. For each language, I show the plots colored by (1) the Dependent-First logit for the language, (2) the Distance logit, (3) the fixed effects that were included the model, among DepL (dependence length), surprisal, and ID (Variance of Surprisal (for UID)).

In general, slopes tend to become weaker when controlling for surprisal than when dependency length is the only fixed effect.

## 
## ###########
## Korean

## 
## ###########
## Hindi

## 
## ###########
## Japanese

## 
## ###########
## Czech

## 
## ###########
## Russian

## 
## ###########
## English

## 
## ###########
## German

## 
## ###########
## Spanish

## 
## ###########
## Norwegian

Across Languages